We live with technology. Algorithms are the heart and soul of technological experiences that are offered through products, services and systems. With the evolution of Internet and ever increasing volume and variety of data, algorithms are now designed and implemented to analyze, interpret and bring forth meaningful patterns across data.
We live in a world where data is currency. And meaningful data is invaluable. The algorithms designed today are intelligent and capable enough to be predictive about user behavior, their attitudes, desires, fears and motivations. Algorithms continue to evolve, become more knowledgeable by understanding the data that is being fed to them over time.
As a very valuable analytics tool, algorithms are built to make decisions which may inadvertently be biased or erroneous. Organizations continue to design algorithms for analytics and let the algorithms make automated decisions or share these inputs to stakeholders who can make decisions.
Openness and transparency of algorithm is important as organizations go about using data of large amounts and make decisions. Can organizations take the responsibility of being transparent and making the meaning and the purpose of their product’s algorithm available to all the stakeholders? Some of the questions that come to my mind when I think of transparency of algorithms are -
- What if the algorithm is proprietary (their USP, may be) to an organization? Would they still share what and how the algorithm works? What are the implications of such products when the stakeholders’ unwillingness to share jeopardizes or has a social impact?
- What if the algorithms are made available to public in an easily understandable manner? Would the stakeholders hold themselves accountable when the algorithm has potential violations or implications?
- Can any one affected by the nature and function of the algorithm raise concerns and would the stakeholders be willing to correct or redress the issues?
For users who are not technically savvy enough to understand algorithms and question the nature of it, the concept of algorithm becomes a black box and the users do not have the slightest clue of how the algorithm takes advantage of the users’ data, behaviors, choices and decisions. It makes me wonder if the algorithm defines as to “who we are” and “what we should do or not do”.
One of the principles stated in the “Statement on Algorithmic Transparency and Accountability” article was “Explanation” – As a designer and a user of technology products, what this means to me is an understanding of how the algorithm works and what decisions it makes on behalf of me or what decisions it lets me make when I use such products. Understanding of algorithms is a valuable knowledge and may open up an opportunity for individuals to critique about the nature and the way it works and possibly inspire to create their own interpretations of the algorithm or spark a new idea and also create trust and credibility in the product. If only we knew what happens to our personal data every time we select “Allow” or “Accept” when a new app or game or an update is downloaded on our mobile phones.
I wanted to share an interesting example of Algorithm transparency and data privacy in Evernote’s product called Context. This is a classic example of privacy of data and how algorithms read and interpret individuals’ thoughts as they type it in Evernote. The below except sets the context about the data privacy and the capability of the algorithm in interpreting user data as they type and bring in third party content relevant to what they are typing!
To build Context, Evernote needed a way to match what a user was working on with relevant content. To do that, they hired a team of augmented intelligence researchers, who created an algorithm that is constantly trying to match keywords from what users are typing to related terms in outside material. As you type a note, Evernote not only searches for and provides relevant material from your own and colleague’s work, but also from third-party news organizations.
To get this material, Evernote had to locate willing media partners who would give the company access to some or all of the news organization’s archives. Their first partner was The Wall Street Journal, which makes the the most recent 90 days of its archives available and searchable via Context. He believes that if people better understood how Context works — that the data flows only one way and that there’s no money involved — they would be more open to the benefits of the product. Without that information being made explicit, users often assume the worst of tech companies;
You can continue reading this article at this link:
http://www.niemanlab.org/2014/12/algorithm-fatigue-what-evernotes-news-recommending-product-can-tell-us-about-privacy/
PROJECT 1
